https://pantsbuild.org/ logo
Join Slack
Powered by
Is it expected behavior that listing the transitiv...
# general
c
Is it expected behavior that listing the transitive dependencies of a PEX target list a few package, but misses many from the actual PEX? Shouldn't the transitive dependencies include package dependencies? I wish I could just use 
pants paths ...
to figure out why I have 
mypy
and 
grpcio
in my PEX package
g
You should be able to list all targets with 
pants list ::
and then identify the target path to 
mypy
and 
grpcio
and then use 
pants dependents $grpcio_target
That might help identify what depends on it. Then you can use 
pants paths --from=$discovered_dep --to=$grpcio_target
and do a for loop over the discovered deps
h
That works for the first level of deps the pex_binary target depends on, but not the transitive deps that are resolved by pip (and saved in the lockfile)
Currently those are not modeled in Pants as first-class targets
There is discussion of this here: https://github.com/pantsbuild/pants/issues/12733
c
Thanks Benjy, this is what I was refering to!
Since Pants doesn't infer dependencies dependencies, how does it deal with mypy? We seems to have issues due to this, where 
opensearch-py
depends on 
six
, but not our project, so Mypy throw some error about 
types-six
missing while parsing 
opensearch-py
, is this a known issue?
Look more like a race condition like in https://github.com/pantsbuild/pants/issues/18519
4 Views
Open in Slack
PreviousNext
Powered by